Sawmill Production Planning
If commerce sets the headroom, planning decides how much of it you actually capture. In our view planning is a high-leverage lever and often an underbuilt one, because it tends to live in spreadsheets and one technologist's head. This article explains why short-horizon planning constrains the planner, what changes when you plan on a window, and what good sawmill optimization looks like in practice — built around sawing batch planning and an explicit yield model rather than rules of thumb. Numbers are labelled by origin: cited from a source, SawmillSmart's own estimate, or illustrative.
Why planning is the highest-leverage lever
Commerce sets the menu the planner has to work with. Sorting decides whether the planner's decisions are physically executable. Sawing, drying, and dry sorting execute or destroy the plan. Planning sits in the middle of the chain, and in our view it is the highest-leverage step for lumber yield improvement because it is the only step that chooses. Every other step either prepares the input for that choice or carries out its consequences. "Highest-leverage" is our judgement, not a measured ranking of steps. The research literature does describe sawmill production planning as a key factor for the Argentine sawmill sector (Vanzetti et al., 2018) and as challenging, among other things because of the heterogeneity of the raw material (Chabanet et al., 2022).
Many sawmills under-invest in planning for a specific reason: planning doesn't look like work. The salesperson is on the phone. The saw is loud and visible. The kiln has temperature readouts and clear cycle times. The technologist is at a desk with a spreadsheet, deciding which cant pattern goes on the line tomorrow. From the outside, that looks like paperwork. In reality, that choice decides which boards the line will produce and what they can be sold for.
The thesis of this article is simple: planning one shift at a time, with the order book that happens to be open, against rules of thumb the technologist has built up over years, can leave yield and order coverage on the table. How much depends on the mill, and we do not claim a universal figure. The fix we propose is structural, not personal. Replace the shift-at-a-time horizon with a planning window. Replace the closed order book with an order pool of roughly 1.5× the production volume (SawmillSmart's rule of thumb, explained below). Replace rules of thumb with a yield model that quantifies the trade-offs.
The rest of this article walks through each piece of that fix.
The shift trap — why mills default to short-horizon planning
Before getting to what horizon planning is, it's worth understanding why many mills don't do it. The short answer: shift-by-shift planning feels safer, and it maps to how the rest of the mill is organized. In our view it usually leaves value on the table. The three forces and four costs below are SawmillSmart's observations from project work, not survey results.
Why shift planning is so common
Three forces push mills toward the short horizon:
The information the technologist has on Monday morning feels concrete. Today's sorted log buffer is what's in front of them. Today's open orders are in the order book. Today's saw line is set up the way it was on Friday. Tomorrow is more uncertain — a customer might push out, a saw might need an alignment pause, a log delivery might delay. So the technologist plans the day they can see and trusts they'll figure out the next day when it arrives.
Reactive culture rewards it. When a customer calls Friday afternoon with an urgent order, the response everyone respects is "we'll find a way to fit it in." That response is only possible if the plan is short. A two-week plan that says "we can fit you in week 3" feels less helpful than "we'll figure something out tomorrow," even when the two-week plan would deliver better economics for both sides.
Tooling enforces it. Spreadsheets are great for a single day's plan and miserable for a two-week rolling plan. ERPs typically record what happened rather than propose what should happen next. Without dedicated planning software, the friction of planning a longer horizon is so high that no one does it.
The hidden costs of shift planning
The visible cost of shift planning is none — the mill keeps running. The hidden cost is yield and order coverage left on the floor, and it can show up in four ways:
- Forced fits. With one shift of orders and one shift of sorted logs, the planner has only a few cant pattern options. The chosen pattern is the least-bad option in a small set. Yield is whatever that least-bad option produces.
- Order whiplash. Orders that the planner couldn't fit today get dropped into tomorrow with no context. Tomorrow's planner — sometimes the same person, sometimes a different shift's technologist — starts from scratch.
- Kiln misalignment. Drying is a major processing step in sawmills with high production volumes (Huka et al., 2020), and in SawmillSmart's observation the kiln pipeline runs in days, not hours. A planner working on today's saw line can't see the kiln charges that will need feeding next week. Half-empty charges can become routine.
- Customer-service trade-offs. Without a horizon, every customer request is a binary: yes today, or no until further notice. Mills with a horizon can offer "we'll deliver in week 3" with confidence — and some customers may prefer that to a rushed commitment.
The horizon principle — what changes when you plan on a window
Horizon planning is exactly what it sounds like: instead of optimizing one shift at a time, the planner optimizes a window — for example one to two weeks of production — as a single problem. Every shift in the window is still planned; the difference is that each shift's plan is chosen with knowledge of every other shift's plan, the full sorted log inventory across diameters, the full order pool, and the kiln pipeline downstream. This is the foundation of effective sawing batch planning: not "what runs next?" but "what runs across the window, and in what sequence?"
The Toyota parallel — why this idea is not new
The horizon principle is not a sawmill invention. A well-known version of the underlying thinking is the Toyota Production System (TPS), developed at Toyota under Taiichi Ohno; Sugimori et al. (1977) describe it as having taken root at Toyota over the preceding 20 years. TPS addressed a problem that looks familiar to a sawmill: making only the necessary products, at the necessary time and in the necessary quantity, while holding stock on hand to a minimum (Sugimori et al.; Toyota Motor Corporation) — although the equipment and the raw material are different.
Three TPS ideas offer useful analogies for sawmill optimization (the mapping is our interpretation):
- Heijunka — production smoothing. Toyota's practice of smoothed production is described in Monden (1994). In sawmill terms the closest analogue is the planning horizon: treat the window as one problem. The analogy is ours; a sawmill smooths a different thing.
- Just-in-Time: only what is needed, when it is needed, in the amount needed. Toyota describes Just-in-Time this way (Toyota Motor Corporation), and Sugimori et al. add that stock on hand is held to a minimum. The order pool raises a related trade-off for sawmills: too small an order pool and the planner has no choice; too large and working capital balloons. Calling the 1.5× order pool a "JIT buffer" is our analogy, not a Toyota practice.
- Andon — surface problems immediately. At Toyota the andon board lights up when equipment stops or when a worker pulls the stop cord, so that the person in charge can respond (Toyota Motor Corporation). A mill that reviews plan versus actual only at month-end hears about a miss long after it happened. Comparing planned and actual yield the same day, while the cause can still be investigated, is the andon idea applied to lumber yield improvement.
The analogy is not perfect — Toyota assembles standardized parts into standardized products, while a sawmill receives variable raw input (logs) and produces several products at once from the same log: a divergent process with co-production (Dumetz et al., 2016), where raw material heterogeneity and divergent product flows make operations difficult to manage (Dumetz et al., 2015). The underlying ideas — a wider planning horizon, a right-sized buffer, quick feedback — are what we borrow. This article does not claim a Toyota-sized improvement for sawmills.
The next four sections walk through four mechanisms through which a longer horizon can help in a sawmill. They are SawmillSmart's description, and the examples use illustrative numbers.
Mechanism 1 — Pattern choice freedom
With one shift of inventory and one shift of orders, the planner may have only a few viable cant patterns to choose from. With a one- to two-week horizon, the candidates multiply across all available diameters and all open orders, and the planner can pick the handful that will run during the window. The counts in the diagram (2 versus 12–15) are illustrative.
The best option in a larger set is never worse than the best option in a subset of it; how much better depends on the mill, its logs and its orders. A planner working from a small set gets the least-bad option. A planner working from a large set can get a genuinely good one. Published sawmill planning models make the same move: they select the set of cutting patterns for each planning day at once, against the required demand (Vanzetti et al., 2018).
Mechanism 2 — Inventory timing
Sorted log inventory shifts day by day. A diameter that's well-stocked Monday may run dry Wednesday and refill by Friday. A planner working on today's shift takes whatever is in front of them and cuts. A planner working on a horizon can wait — assign a particular set of orders to Thursday's shift because by then the 200 mm bin will be full, while running Monday's shift on 180 mm.
The savings here are not yield-per-cut savings; they are option-value savings. The horizon lets the planner spend log inventory at the moment when its match to orders is best, rather than at the moment it's available.
Mechanism 3 — Cross-shift continuity
Shift planning treats each shift as independent. An order that was 60% cleared at the end of Monday gets dropped into Tuesday with no continuity. Tuesday's planner — possibly the next-shift technologist — starts the cant pattern decision from scratch, often choosing differently. The remaining 40% of Monday's order can take more setups and more changeovers than it needs to (illustrative percentages).
Horizon planning carries continuity across shifts. The same cant pattern runs Monday afternoon into Tuesday morning to clear the open order; the changeover happens once, at the end of the order rather than at the end of each shift.
Mechanism 4 — Kiln pipeline alignment
Kiln drying runs on its own cycle, and boards cut Monday do not finish drying on Monday. Kiln loading and scheduling is a planning problem of its own (Huka et al., 2020), and a recent multiperiod planning study frames drying as a frequent bottleneck for lead times and for meeting customer demand (Kökten and Bayraktar, 2026). A planner working on today's shift can't see which kiln charges will be ready for filling next Tuesday, so charges can end up assembled reactively from whatever boards happen to come off the line.
Horizon planning aligns the cut to the dry. A planner who can see that next Tuesday's kiln charge needs 22 mm boards can plan this Monday's cant pattern to produce 22 mm boards in the right volume — turning the kiln from a passive consumer into an active partner in the plan. How much capacity this recovers depends on the mill; we have no independent figure for it.
The 1.5× order pool — sizing the input
Horizon planning only works if there's something to choose from. A 2-week horizon over an order book exactly equal to 2 weeks of production volume still produces forced fits — every order has to go somewhere, and the optimization problem is one with no slack. To unlock the value of the horizon, the order pool needs to be larger than the production volume of the window. By order pool we mean open, committed orders, measured in m³ of sawn timber and compared with the timber the line will produce in the same window.
SawmillSmart's working rule of thumb is roughly 1.5× the production volume of the planning window. This is a heuristic from our project work, not a value derived from a published study; we found no independent source for an optimal pool ratio, and the right number will differ by mill. As an illustrative calculation: for a two-week window producing 4 000 m³ of timber, a 1.5× pool means 4 000 × 1.5 = 6 000 m³ of open, committed orders, eligible to be cut from that production.
Why 1.5× and not 1.0× or 2.0×
The bands below are rules of thumb. The thresholds are not taken from a study; they describe how the trade-off tends to look in SawmillSmart's project work.
Below 1.3× — forced fits. Every order in the book has to go through the line in the planning window. The planner has no slack. The optimization problem degenerates back to the shift trap, just on a longer horizon.
1.3× to 1.5× — workable. The planner has some choice, particularly for high-volume specs that overlap multiple diameters. Yield and order coverage can improve. In our rule of thumb this is the minimum workable pool.
1.5× — working target. Pattern choice is genuinely open. The planner can balance yield, order coverage, and kiln pipeline simultaneously. Working capital tied up in promised-but-uncut orders is manageable. Customer service is easier to preserve because most orders can clear within their commit window.
1.7× to 2.0× — diminishing returns plus real costs. The marginal yield gain from a larger pool can flatten out — there are only so many patterns that ever get chosen, no matter how many candidates exist. Meanwhile, working capital tied up in promised orders grows, customers wait longer than they should, and the long tail of small orders in the pool starts to look more like overbooking than optimization.
How to grow the pool from 1.0× to 1.5×
For some mills the pool starts well below 1.0× — sales books orders just-in-time, and the order book at any moment barely covers the next week of production. Growing the pool to 1.5× takes deliberate commercial action. Three practices can help:
- Forward-sell with delivery flexibility. Offer a delivery window instead of a fixed date, for example 2–3 weeks. Some customers may accept the window in exchange for slightly better pricing or priority access.
- Recruit complementary orders, not duplicates. Sales should be looking for orders that complement what's already in the book — thin specs to match dominant thick orders, backfill orders for under-subscribed kiln charges (see Practice 1B in the Commerce article).
- Build relationships with anchor customers willing to commit on a window. A short list of customers who commit volume on a 2-week window — not a fixed date — gives the planner the slack they need. These customers may expect preferential pricing or service in exchange.
The yield model — yield × order coverage, not yield alone
A larger pool of orders gives the planner more options. The yield model is what tells the planner which option is best. And "best" turns out to be different from what a yield-only model implicitly assumes.
A common planning instinct — and what a yield-only calculator encodes — is to maximize yield percentage: of the candidate cant patterns for a given log diameter, pick the one that produces the highest cubic meters of timber per cubic meter of log. When the output also has to match open orders, this objective can mislead, and the failure mode is specific.
The yield-only trap
Hypothetical: a cant pattern with 53% theoretical yield that produces boards no current customer wants can be worse than a pattern with 51% theoretical yield that clears two open orders.
The 53% pattern produces 2 pp more timber per log. The 51% pattern produces 2 pp less timber per log — but every board has a buyer at the contract price. The 53% pattern's "extra" boards go to stock (working capital tied up) or to commodity sale at a lower price per m³. Whether the 51% pattern wins depends on that price gap; the worked example below shows the arithmetic and the break-even.
The right objective function is yield × order coverage — the cubic meters of timber that clear an open order at standard price per cubic meter of log. Equivalently: revenue per m³ of log, calculated at the price the order actually pays, not at the price the boards would fetch in the spot market.
Research on sawing optimization points in the same direction. Todoroki and Rönnqvist (2002) linked log sawing optimization to production against an order book, with product values that were updated regularly according to previous timber production. On three order books, tested by simulation on pruned Pinus radiata logs, product optimization needed fewer logs to satisfy all demands than volume or value optimization, eliminated underproduction and reduced overproduction. It is a simulation study, not a mill trial, and it does not measure the € effect.
This shifts every decision. Patterns are now ranked by their expected revenue contribution to the order book in the planning window — not by their abstract yield. Pattern selection respects what's open, not what's possible in theory. The technologist still chooses, but they choose with the right number in front of them.
Worked example — Monday morning at the mill
Put all three pieces — horizon, pool, yield model — together into a single Monday-morning sawing batch planning scenario. This is an illustrative example: all numbers are assumptions chosen to show the structure of the decision, they are not measured data and are not drawn from a specific mill.
The setup
Production target this week: 2 000 m³ of log throughput (about 1 000 m³ of sawn timber at roughly 50% yield), single-line softwood mill.
Sorted log inventory available:
| Diameter class | Volume in buffer | Replenishment expected |
|---|---|---|
| 160 mm | 600 m³ | steady |
| 180 mm | 800 m³ | peak, draws down by Thursday |
| 200 mm | 600 m³ | refill Wed–Fri |
Open orders in the book this Monday morning:
| Spec | Outstanding volume | Delivery commit |
|---|---|---|
| 22×100 | 400 m³ | this week (priority) |
| 25×100 | 200 m³ | this week |
| 32×100 | 150 m³ | this week |
| 47×100 | 250 m³ | this week |
| 50×150 | 100 m³ | this week |
Total open orders this week: 1 100 m³ of sawn timber. Expected timber production: about 1 000 m³. Order pool / production = 1 100 / 1 000 = 1.1×. Too tight to optimize.
Pool 1.1× — what the planner has to do
With essentially no slack in the pool, every order has to be cut this week. The planner picks the cant pattern with the highest yield for each diameter class and lets the orders fall where they fall.
For 180 mm logs, the highest-yield pattern available is Pattern A — 52% net yield, with 40% of its output matching standing orders. The planner picks A.
Result: the boards of Pattern A that match no standing order — high-yield but not matched to the priority spec — accumulate as stock or sell at the commodity price. The technologist hits the yield number but the commercial team is on the phone all week explaining the partial delivery on 22×100.
Pool 1.5× — what changes
Now imagine sales has spent the past two weeks recruiting complementary orders into the pool. The pool grows by 400 m³ of new orders — sized to complement the existing book, not duplicate it:
- 22×125 — 150 m³ (deliverable in 2 weeks)
- 22×75 — 100 m³ (backfill for kiln charge)
- 32×150 — 100 m³ (deliverable in 2 weeks)
- 50×100 — 50 m³ (deliverable in 2 weeks)
Total order pool: 1 500 m³ of sawn timber. Expected timber production: about 1 000 m³. Order pool / production = 1 500 / 1 000 = 1.5×. The planner now has genuine choice.
For 180 mm logs, a planning tool that ranks patterns against the order pool (as SawmillSmart does) might show three candidate cant patterns, with assumed values:
| Pattern | Net yield | Share of pattern output going to standing orders (assumed) |
|---|---|---|
| A | 52% | 22×100 (40% of output) — rest unsold |
| B | 51% | 22×100 (60% of output) — rest unsold |
| C | 49% | 22×100 (60% of output) + 22×125 (25%) + 22×75 (15%) — 100% to open orders |
With pool 1.1×, the planner picked A. Yield 52%, but coverage was a problem.
With pool 1.5×, the planner picks C. The pattern delivers 49% yield — 3 percentage points below A — but every board it produces clears a standing order at contract price. There's no commodity fall-down. The 22×100 priority order receives more of its volume from the same logs (60% of the pattern's output instead of 40%), and the 22×125 and 22×75 backfill orders progress in parallel.
The economic math
Crucially, the planner has not lost yield value. They have traded headline yield for revenue per m³. Walk the math with assumed prices — €280/m³ as the standing-order contract price and €240/m³ as the commodity fall-down price — on 1 000 m³ of logs:
| Metric | Pattern A (yield-only) | Pattern C (yield × coverage) |
|---|---|---|
| Net yield | 52% | 49% |
| Timber from 1 000 m³ of logs | 520 m³ | 490 m³ |
| Volume to standing orders | 208 m³ (40%) | 490 m³ (100%) |
| Volume to commodity | 312 m³ (60%) | 0 m³ |
| Revenue at €280 / €240 per m³ | €133 120 | €137 200 |
| Effective price per m³ of timber | €256 | €280 |
| Revenue per m³ of log input | €133 | €137 |
Formulas. Pattern A: timber = 1 000 × 0.52 = 520 m³; to standing orders = 520 × 0.40 = 208 m³; to commodity = 520 − 208 = 312 m³; revenue = 208 × €280 + 312 × €240 = €58 240 + €74 880 = €133 120. Pattern C: timber = 1 000 × 0.49 = 490 m³, all to standing orders; revenue = 490 × €280 = €137 200. Per m³ of log input: €133.12 versus €137.20, a difference of €4.08.
Under these assumptions Pattern C produces about €4 more revenue per m³ of log input than Pattern A, while looking — to anyone who watches only the yield column — like the worse pattern. At 2 000 m³ of logs per week, €4.08 × 2 000 = €8 160 per week, or about €408 000 per year over 50 operating weeks. The result depends entirely on the assumed price gap and on the assumed share of Pattern A's output that finds no standing order: Pattern A wins if the commodity price is above €253/m³ (break-even: 208 × 280 + 312 × p = 137 200, so p = €253.08). Run the same arithmetic with your own prices before drawing conclusions. In this example only the planning logic differs: same logs, same line.
This is the case for the yield model in one example. The shift planner with pool 1.1× picks A. The horizon planner with pool 1.5× and a yield-times-coverage model picks C. Same data, different objective, different answer.
The anonymized case — yield, coverage, and revenue together
The worked example above is illustrative. The case below is a real lumber yield improvement project, anonymized to preserve commercial confidentiality.
A medium-size European softwood sawmill — approximately 140 000 m³ of annual log intake, mixed pine and spruce, single-line production with a multi-rip primary breakdown and an in-line splitter — moved from spreadsheet-based shift planning to horizon planning with a 1.5× order pool and a yield-coverage model. The before/after data is from two consecutive quarters with comparable log mix and order book composition.
What the three metrics tell you together
Most planning conversations focus on yield alone because it's the easiest number to track. The case above shows why looking at yield alone misses the larger story.
Yield rose 5 percentage points. From 46% to 51%. This is the headline number — same logs producing more timber. Worth approximately €1.4M/year in additional timber output at average market price.
Order coverage rose 24 percentage points. From 64% to 88%. The mill went from selling roughly two-thirds of its timber against open contracts to selling almost all of it. The other 36% (before) was going to stock buildup or commodity sale at discount. This shift alone — the same volume of timber, but a higher share into standing orders — was worth approximately €400 K/year in revenue mix uplift.
Average revenue per m³ of timber rose €18. From €247 to €265. This is the consequence of the order coverage improvement: timber that lands on a contract is paid at contract price, while timber that doesn't lands in commodity at €30–50 less per m³. The shift in revenue per m³ compounds with the yield gain to produce the headline annual margin impact.
Together, these three metrics tell the full story that yield alone hides. The mill produced more timber, sold a larger share of it at standing-order prices, and earned more per m³ of what it sold. None of those three is the cause of the others — they are all downstream consequences of the same upstream change: better planning.
How SawmillSmart implements horizon planning
Everything above is the principle. SawmillSmart Planning is one implementation of it. This section describes only what SawmillSmart publishes on its product pages (product overview, how it works) and makes no claim beyond them.
SawmillSmart is production-planning software for sawmills. It optimizes production planning across a multi-week horizon — typically several weeks — evaluating thousands of cutting combinations, and it keeps a rolling four-week plan. It needs no new equipment on the line.
Practice 1 — Cutting patterns evaluated against your order book
How it works. The planner loads the inputs the mill already keeps: the log assortment (species, diameter, length, count), the orders or a demand plan, and the sawline capacities, sawing parameters and shift structure. Orders can be imported from Excel. SawmillSmart then evaluates feasible combinations of cut patterns, production allocation and sequencing under all defined constraints at once, and selects the combination that maximizes recovery while ensuring demand fulfillment.
What's different. A spreadsheet yield calculator can tell you the theoretical yield of one pattern on one log diameter. It can't tell you whether the boards that pattern produces match an open order. SawmillSmart ranks simulated scenarios by lumber recovery, log cost and fulfillment match, so the question for this set of logs and this set of orders, which patterns are worth running? is answered in one place.
Output the planner sees. A ranked list of scenarios with a full economic breakdown — m³ processed, recovery %, value per log and total production cost — which can be compared side by side. The technologist still chooses.
Practice 2 — Recovery visible before the cut
What it does. Each simulation run shows recovery % before anything is committed to the line. For every cutting pattern in a simulation, SawmillSmart renders a cross-section view against the actual log diameter — board placement, wane allowance zones and effective zone boundaries — with yield percentage, board count and log volume alongside.
Why this matters. Without a figure in front of them, every cant pattern is a bet: the technologist commits the pattern, the line runs it, and the actual yield shows up at the end of the shift, too late to choose differently. With a simulated figure the planner can compare patterns before committing. A simulation is only as good as its inputs: SawmillSmart's own materials state that the quality of the inputs determines the quality of the plan.
Practice 3 — A rolling plan that updates
What it does. The rolling planner maintains a plan across the next four weeks and updates it as orders change, inventory shifts or line conditions evolve. The operational plan view shows order fulfillment progress, frozen-zone protection for committed shifts, plan coverage % and at-risk order alerts, and the shift schedule is generated from the confirmed plan.
Why it matters. Orders, logs and line conditions change inside a window, and a horizon plan that cannot be revised stops being a plan. Whether a given mill gets the gains it hopes for depends on its data, its constraints and its order book, so SawmillSmart's own offer is to run a simulation on the mill's real data in a demo.
What SawmillSmart's own figures are
SawmillSmart's product page cites a 2–4% average yield improvement versus shift-by-shift planning. It is SawmillSmart's own estimate, not an independent study, and the method is not published there. The product page does not say whether the figure is percent or percentage points. The case above is anonymized SawmillSmart customer data from a single mill. For your mill, the only meaningful figure is one computed on your own data.
What planning is NOT — three anti-patterns
Conversations about sawmill planning often confuse three distinct functions. Each of them lives next to planning in the workflow, and each is sometimes called planning by people who use the word loosely. None of them is what this article is about, and conflating them can be one reason mills under-invest in actual planning.
Anti-pattern 1 — Planning is not scheduling
Scheduling answers the question "what runs when?" — Monday morning shift, Tuesday afternoon shift, Friday night. Scheduling is a calendar exercise. It cares about resource availability, shift coverage, and maintenance windows.
Planning answers a different question: "what should we cut, from which logs, against which orders?" — a yield-and-coverage exercise. Planning's output feeds the schedule, but the schedule doesn't optimize anything by itself. A perfectly scheduled mill running sub-optimal cant patterns produces sub-optimal yield with great timing.
In our view, buying a better scheduler does not by itself fix the planning gap.
Anti-pattern 2 — Planning is not queueing
Queueing is the order in which open orders get cut. First-in-first-out. Largest-first. By delivery date. By customer priority. These are queueing rules, and most mills have them — explicit or implicit.
Queueing rules are downstream of planning. The planner decides which cant patterns to run; queueing only matters once that decision is made, to assign output to specific orders. A great queueing system applied to a bad cant pattern decision produces orderly delivery of yield-poor cuts.
Anti-pattern 3 — Planning is not order entry
Order entry is the workflow of getting a customer order into the system: spec, volume, delivery, price. It's a commercial function (see sawmill commerce).
The order book is the input to planning, not planning itself. A clean order book is a precondition for good planning; it doesn't replace it. Better order entry makes the input cleaner; it does not change the optimization step that consumes it.
What to do this week — checklist
Five diagnostic actions that surface where the planning gap may be at your mill and what to do about it. The thresholds are SawmillSmart rules of thumb for triage, not validated benchmarks.
Sawmill planning — week 1 checklist
- Calculate your order pool ratio. Total volume of open orders ÷ production volume in the planning window, both in m³ of sawn timber. Below about 1.3× there is little room to choose, whatever the tooling. Above about 1.7× the pool may be too big and customer service at risk.
- Audit one week of plan vs. actual yield. Pull the planned yield for each shift in the past week and compare to actual. A gap of more than about 2 percentage points suggests the plan isn't anchored in a real yield model.
- Pick one open order and trace it through last week. For your single highest-priority open order, count how many separate cant patterns it appeared in across the week. More than about 3 suggests continuity is broken; the order is being re-planned every shift.
- Walk to the kiln. Count under-loaded charges in the last week. Each under-loaded charge may be downstream evidence of planning that didn't see the kiln pipeline.
- Ask the technologist what tooling they actually use to plan. If the answer is "Excel" or "we always run pattern X for diameter Y" — there may be no yield model in the loop. That's the planning gap.
Where to go next
If two or more of the checklist items above came back red, the planning gap may be worth quantifying directly. In a demo, SawmillSmart walks through your order structure and shows you a real simulation run.
Sources & further reading
Each source below was checked against its Crossref record. The statements we rely on come from the abstract (read via OpenAlex), from the open-access full text (Vanzetti et al., 2018) or from the page itself (Toyota); the Monden chapter was checked by title and metadata only, its abstract was not opened. What is not sourced: the 1.5× order pool and its bands (SawmillSmart's rule of thumb), the worked example (assumed numbers), the anonymized case from a real mill (SawmillSmart customer data, a single mill) and the 2–4% estimate (SawmillSmart's own).
Sawmill planning and sawing optimization
- Todoroki, C.L., & Rönnqvist, M. (2002). Dynamic control of timber production at a sawmill with log sawing optimization. Scandinavian Journal of Forest Research, 17(1), 79–89. doi:10.1080/028275802317221118
- Vanzetti, N., Broz, D., Corsano, G., & Montagna, J.M. (2018). An optimization approach for multiperiod production planning in a sawmill. Forest Policy and Economics, 97, 1–8. doi:10.1016/j.forpol.2018.09.001
- Chabanet, S., Bril El-Haouzi, H., Morin, M., Gaudreault, J., & Thomas, P. (2022). Toward digital twins for sawmill production planning and control: benefits, opportunities, and challenges. International Journal of Production Research, 61(7), 2190–2213. doi:10.1080/00207543.2022.2068086
- Dumetz, L., Gaudreault, J., Thomas, A., Marier, P., Lehoux, N., & El-Haouzi, H. (2015). A simulation framework for the evaluation of production planning and order management strategies in the sawmilling industry. IFAC-PapersOnLine, 48(3), 622–627. doi:10.1016/j.ifacol.2015.06.151
- Dumetz, L., Gaudreault, J., Thomas, A., Lehoux, N., Marier, P., & El-Haouzi, H. (2016). Evaluating order acceptance policies for divergent production systems with co-production. International Journal of Production Research, 55(13), 3631–3643. doi:10.1080/00207543.2016.1193250
Kiln drying and planning
- Huka, M.A., Rindler, C., & Gronalt, M. (2020). Scheduling and loading problem for multiple, identical dry kilns. Flexible Services and Manufacturing Journal, 33(2), 312–336. doi:10.1007/s10696-020-09386-4
- Kökten, E.S., & Bayraktar, T. (2026). Integrated production and drying-capacity planning in sawmilling: a multi-period mixed integer linear programming optimization approach. Forest Products Journal, 76(2), 183–194. doi:10.13073/fpj-d-26-00007
Toyota Production System
- Sugimori, Y., Kusunoki, K., Cho, F., & Uchikawa, S. (1977). Toyota production system and Kanban system: materialization of just-in-time and respect-for-human system. International Journal of Production Research, 15(6), 553–564. doi:10.1080/00207547708943149
- Monden, Y. (1994). Smoothed production helps Toyota adapt to demand changes and reduce inventory. In: Toyota Production System, Springer, 63–73. doi:10.1007/978-1-4615-9714-8_4
- Toyota Motor Corporation. Toyota Production System (company page on Just-in-Time, jidoka and andon; accessed 2026-10-23). global.toyota
The case (yield 46% → 51%, order coverage 64% → 88%, revenue +€18/m³) is anonymized SawmillSmart customer data from one real European softwood mill. The Monday-morning worked example uses assumed numbers for illustration and is not drawn from a specific mill.
